https://doi.org/10.1007/s00024-018-2027-2 2
3
Quantifying the partition between seismic and aseismic deformation along 4
creeping and locked sections of the North Anatolian Fault, Turkey 5 6
Maor Kaduri1, Jean-Pierre Gratier1, Cécile Lasserre2, Ziyadin Çakir3, François Renard1,4,*
7 8
1 Université Grenoble Alpes, Université Savoie Mont Blanc, CNRS, IRD, IFSTTAR, ISTerre, 38000 9
Grenoble, France 10
2 Université de Lyon, UCBL, ENSL, CNRS, LGL-TPE, 69622 Villeurbanne, France 11
3 Istanbul Technical University, Department of Geology, Istanbul, Turkey 12
4 The Njord Centre, PGP, Department of Geosciences, University of Oslo, box 1048, 0316 Blindern, 13
Oslo, Norway 14
15
*Correspondence should be addressed to François Renard ([email protected]) 16 17
18
Abstract 19
Shallow aseismic creep is a key deformation component along plate boundaries that 20
contributes to the energy budget during the seismic cycle. Several major active 21
continental faults show spatial alternation of creeping and locked sections. The 22
present study focuses on the evaluation of the aseismic part of the total displacement 23
along the North Anatolian Fault in Turkey. Detailed microstructural analyses of finite 24
strain were performed using various methods, based on change of length or angle, on 25
six representative samples collected over 32 outcrops along locked and creeping 26
sections of the fault. Chemical analyses were used to map mineral composition of 27
fault rocks and to calculate relative volume changes associated with creep.
28
Relationship between finite strain and volume change allowed quantifying the 29
evolution of the penetrative pressure solution cleavage mechanism of creep. In 30
volcanic and analogous creeping rocks, finite strain measurements revealed two 31
spatial scales of strain that correspond to the alternation of two types of shear zones, 32
with cleavages either oblique or sub-parallel to the fault displacement. Using geodetic 33
and geological data, cumulative aseismic displacement was calculated in the range 9 34
to 49% of the total 80 km displacement in the creping sections, and was negligible in 35
locked sections. The large uncertainty in the kilometer-width creeping sections was 36
related to the difficulty of quantifying high strain values that are associated with high 37
shear displacement and for which measurement uncertainties are large. A promising 38
way to improve such quantification would be to develop reliable statistical analysis of 39
cleavage orientation in the field.
40 41
1. Introduction 42
Aseismic creep has been extensively documented along several active faults 43
worldwide (Chen and Bürgmann, 2017; Harris, 2017; Bürgmann, 2018). Active fault 44
creep processes develop either as transient (mostly post-seismic) sliding or as 45
permanent sliding. Post-seismic creep rate can evolve through time with various 46
patterns: with an exponential or a power law decay, down to zero or to a residual 47
constant creep velocity (Çakir et al., 2005). In some cases, shallow creep can 48
accommodate the whole tectonic loading as, for example, along the permanent 49
creeping section of the San Andreas fault (Savage and Burford, 1973). In cases where 50
shallow creep only partly releases the tectonic loading, as is the case along the North 51
Anatolian, the Hayward or the Longitudinal Valley faults in Turkey, California and 52
Taiwan, respectively (Çakir et al., 2005, Graymer et al. 2005, Thomas et al., 2014), 53
major earthquakes may still occur at depth and propagate toward the surface.
54
These contrasting creep behavior patterns may have different implications for seismic 55
hazard assessment. It is therefore crucial to understand the spatio-temporal 56
characteristics of creep and its mechanisms (Bürgmann, 2018). Rock types can partly 57
control fault creep behavior. For example, observations have shown that surface creep 58
rates along the North Anatolian Fault are almost null along fault sections that are rich 59
in massive limestones, but become significant along sections that host various kinds 60
of volcanic rocks that were softened by progressive deformation processes (Kaduri et 61
al., 2017). However, other important characteristics of creep, such as creep 62
deformation duration during a single seismic cycle and over several seismic cycles, as 63
well as the aseismic part of the total displacement during geological fault history, 64
remain to be determined more accurately.
65
To make progress in understanding creep mechanisms, the displacement associated 66
with creep processes must be evaluated. On one hand, the total displacement over the 67
lifetime of a fault accommodated by both creep and earthquakes can be evaluated by 68
measuring the offset of a number of geological markers (Emre et al., 2013). On the 69
other hand, the displacement associated with creep processes can be deduced only 70
from geodetic or geological strain measurements, quantifying the short-term 71
(elastic) parts of the deformation. The distinction between creep and tectonic loading 74
is generally made through simple modeling by considering an elastic half-space 75
surrounding a fault plane (Savage and Burford, 1973; Okada, 1985). In addition, 76
geodetic measurements are representative of slip only in the recent past, covering a 77
few years to several decades. Extending geodetic displacement measurements to the 78
long term (geological times of several millions years) must include an assumption on 79
the variations of the displacement rate with time, which is almost impossible to 80
estimate, especially in areas associating seismic and aseismic processes.
81
Consequently, in order to evaluate only the irreversible part of the large finite strain 82
associated with the creep processes, complementary detailed finite strain 83
measurements in the gouge and damage zone rocks of fault zones are required.
84
The aim of the present study is to quantify the displacement related to aseismic 85
irreversible creep processes in both creeping and apparently, presently locked sections 86
along the North Anatolian Fault in Turkey. To reach this aim, one needs to measure 87
the associated geological finite strain in both the gouge and the damage zone rocks 88
(Fig. 1). In order to calculate this displacement, two types of data are required: the 89
finite strain values and the width of the zone of associated shear deformation 90
(Ramsay, 1980; Ramsay and Graham, 1970). However, the measurement and the 91
interpretation of such data are rather complex because strain and fault width values 92
evolve with time and along the fault strike during deformation, since the beginning of 93
the formation of the fault to the present day. Strike-slip creeping faults accumulate 94
strain in the upper crust by plastic and viscoelastic mechanisms involving both shear 95
deformation and mass transfer, developing what is known as compaction or dilatant 96
shear zones (Ramsay, 1980; Ramsay and Graham, 1970). In such zones, the mineral 97
composition and rock fabric change due to fracturing, mass transfer, fluid circulation, 98
chemical alteration and metamorphism (Gratier et al., 2013, 2011; Steward et al., 99
2000; Imber et al., 2001; Jefferies et al., 2006; Collettini, et al., 2009). Moreover, clay 100
gouge comprising low friction minerals such as montmorillonite and saponite also 101
develop (Carpenter et al., 2016; Kaduri et al., 2017; Lockner et al., 2011; Samuelson 102
and Spiers, 2012). In addition, fault growth is not linear with displacement. Power law 103
relationships between maximum displacement and fault width are often reported 104
(Pennacchioni, 2005; Scholz, 2002). At the outcrop scale, irreversible shear strain 105
gradients generally decrease to zero away from the fault over distances of several 106
centimeters to several kilometers, and sometimes over distances of up to 25 km 107
(Mavko, 1981). The lateral evolution of such strain profiles is often non-linear 108
(Pennacchioni, 2005). The change in shear zone width in time and space thus depends 109
on the deformation mechanisms that combine simple shear, pure shear and volume 110
change and that determine the displacement behavior (Bos and Spiers, 2002; van der 111
Pluijm and Marshak, 2010; Fossen and Cavalcante, 2017). Such heterogeneous shear 112
zones can be classified into three categories based on their width variations:
113
increasing, decreasing or constant in time (Hull, 1988). The shape of displacement 114
profiles across faults depends on whether the deformation process is strain hardening 115
(e.g. deformation diffuses into the host rock) or strain softening (e. g. deformation is 116
localized in narrow zones) (Vitale and Mazzoli, 2008). In particular, strain-softening 117
processes involving simple shear with volume loss are consistent with clay gouge 118
formation (Kaduri et al., 2017).
119
Measuring and interpreting finite strain and the width of the associated creeping shear 120
from geodetic and geological observations are challenging tasks, especially when 121
deformation involves very high strain values. Despite considerable uncertainties 122
associated with both types of observations, the data presented in this study of the 123
North Anatolian fault can be used to characterize: (i) the spatial correlation between 124
the creep-related strain measured by geodetic methods and that measured by 125
geological methods, (ii) the control of rock lithology on aseismic creep mechanisms 126
and temporal evolution, (iii) the aseismic part of the total displacement on the fault.
127
Figure 1. Structural and geological map of the study area along the North Anatolian Fault 129
(NAF), modified from Kaduri et al. (2017). (a) Tectonic setting of study area, located east of 130
Istanbul (blue rectangle with NAF in red), adapted from Emre et al., (2013). EAF: East 131
Anatolian Fault, DST: Dead Sea Transform fault. (b) Geological map of the study area with 132
outcrop locations indicated as blue or red circles for the 32 investigated outcrops. Red circles 133
with names locate the six representative outcrops with strain markers described in the present 134
study: Ta=Taskesti (40°34'54.00"N/ 31°1'60.00"E), Ge=Gerede (40°47'3.00"N/
135
32°6'30.00"E), Is=Ismetpasa (40°51'55.00"N/ 32°35'41.00"E), Ha=Hamamli (40°52'12.19"N/
136
32°39'8.78"E) Ya= Yazioren (40°56'1.59"N/ 33°6'11.09"E), Mu=Mülayim (41°3'5.28"N/
137
33°48'17.61"E). The presently creeping segments of Izmit=Iz (west) and Ismetpasa (east) are 138
marked in white (Çakir et al., 2005; Cetin et al. 2014). The locked segments are marked in 139
black. The regional geological map is extracted from an interactive map of the Turkish 140
geological survey (Akbas et al., 2016) available at
141
http://yerbilimleri.mta.gov.tr/anasayfa.aspx. The legend gives the lithology of the formations 142
that are cut by the studied North Anatolian Fault section.
143 144
2. Measuring creep-related deformation from fault zone outcrops: setting and 145
approach 146
147
2.1 Seismotectonic setting 148
The North Anatolian Fault (NAF) in Turkey is a right-lateral strike-slip plate 149
boundary. This major continental fault accommodates the relative motion between the 150
Anatolian plate, to the south, and the Eurasian plate, to the north. The NAF formed 151
approximately 13 to 11 Ma ago in the east and slowly propagated westward (Şengör, 152
et al. 2004). The long-term geological rate is 20 ±8 mm/yr (Şengör, et al. 2004), 153
consistent with short-term geodetic rates of 24 ±2 mm/yr measured from GPS data 154
(Reilinger et al., 2006) and 25 ±1.5 mm/yr measured from InSAR data (Cetin et al., 155
2014). The total cumulative displacement along the studied NAF segment (Fig. 1) 156
since its initiation is of the order of 80 km (Janssen, et al., 1997; Armijo et al., 1999;
157
Emre et al., 2013; Akbas et al., 2016). A sequence of large earthquakes propagated 158
from east to west and ruptured over ~1000 km of the fault during the 20th century (e.g.
159
Emre et al., 2016), providing evidence of the potential of the fault to accumulate large 160
seismic displacements. However, a striking feature of the NAF is the existence of two 161
slip modes along the fault. Some sections are prone to major earthquakes (Stein et al., 162
1997) and remain locked in between earthquakes while other sections, well-identified 163
by geodesy measurements, display aseismic creep that possibly initiated as post- 164
seismic slip (Kaneko et al., 2013; Çakir et al., 2014; Cetin et al., 2014) and may 165
behave as a succession of transient creep episodes (Bilham et al., 2016; Rousset et al., 166
2016). The relative contribution of seismic and aseismic displacement in the 80 km 167
cumulated displacement remains to be evaluated, which is the aim of the present 168
study.
169
At least two sections of the inland part of the North Anatolian Fault have been 170
identified in recent years as aseismic creeping fault sections (Ismetpasa and Izmit 171
sections, Fig. 1). Based on an InSAR study of Envisat satellite data along descending 172
orbits, it has been shown by Cetin et al. (2014) that the Ismetpasa creeping section 173
runs from longitude 32.2° to 34.0°, with a maximum horizontal surface creep rate of 174
~20 mm/yr, located 30 km east of Ismetpasa, and a creeping depth of ~5 km. These 175
results were revised from a new Envisat InSAR analysis including both descending 176
and ascending data, which helped to obtain a better separation of the vertical and 177
horizontal creep components. The Ismetpasa creeping section was found to be longer 178
(running from 31.0° to 35.0°), with two peaks of maximum horizontal creep rate up to 179
~14 mm/yr, and creep extending down to 9 km depth (Hussain et al., 2016). A clear 180
correlation between shallow creep rate and near-surface fault lithology has been 181
section is composed of volcanic and ophiolite rocks including clay-rich gouges (Fig.
184 185 1).
186
2.2 Fault sampling and strategy analysis 187
We calculated the aseismic part of the total displacement along both locked and 188
creeping zones of the North Anatolian Fault. The finite strain and the creeping width 189
for various lithologies were evaluated by investigating 32 outcrops, over a 400 km- 190
long section between the city of Izmit to the west and the eastern end of the Ismetpasa 191
creeping section to the east (Fig. 1). Satellite images and geological maps (Herece and 192
Akay, 2003, Emre et al., 2013, 2016, Akbaş, et al 2016) were used to identify all 193
accessible outcrops following the procedure described in Kaduri et al., (2017). No 194
continuous outcrop was found that would cover the whole kilometer-width of the 195
NAF shear zone. The links and the common features between these dispersed 196
outcrops is that they all show markers of creep deformation associated with the NAF 197
shear zone, either as penetrative cleavage (in volcanic or analogous rocks) or as 198
stylolites (in limestones). So they all attest of the finite aseismic strain, which was 199
associated with the large strike-slip 80 km-displacement of the NAF shear zone 200
during geological times. It must be noted that geological evidence of such creep 201
deformation was found away from the present-day creeping fault, sometimes at more 202
than one-kilometer distance. This observation indicates that the NAF was not a single 203
localized fault during geological times but a kilometer-width shear zone that include 204
several seismic faults and large associated zones of aseismic deformation, an 205
observation consistent with geological maps (Herece and Akay, 2003, Emre et al., 206
2013, 2016, Akbaş, et al 2016). Building on our first study on the implication of fault 207
rock transformation on aseismic creep (Kaduri et al., 2017), we focus in this 208
following study on the calculation of the total cumulative aseismic displacement from 209
microstructural observations. The main challenge was not only to find outcrops but 210
also to find samples that allow calculating strain values. From the 32 visited outcrops 211
(7 outcrops in the locked segments, 25 outcrops in the creeping segment), 90 oriented 212
hand samples were collected and analyzed in the laboratory and 130 thin polished 213
sections of rock samples were prepared for mineralogical, geochemical and 214
microstructural analyses. All outcrops and samples were chosen to be representative 215
of the wide NAF shear zone and of the geological deformation associated with the 216
NAF displacement. However, estimating strain values was not possible for all these 217
outcrops and samples. Six representative outcrops with strain markers are presented 218
here, one from the locked section and five from the creeping section (Fig. 1). They 219
represent the various aspects of the strain measurement methods and give a 220
representative focus on the evolution from initial rock to damaged rock and gouges, in 221
order to characterize strain related to aseismic creep.
222
Strain was measured at various scales from thin section to outcrop scales. Micro- to 223
meso-structural strain measurements were performed on thin sections from millimeter 224
to decimeter scales. Chemical maps were acquired using either X-ray fluorescence 225
spectrometer (XRF, Eagle III) (mm-cm) or Wavelength Dispersive Spectroscopy on 226
an Electron Probe Micro Analyzer (WDS-EPMA) (microns-mm) (Kaduri et al., 227
2017). Image processing techniques based on MATLAB codes were implemented to 228
calculate internal finite strain from particle distributions or deformation of geological 229
objects. Meso- to macro-structural strain measurements at meter to kilometer scales 230
were derived from field observations and measurements in quarries with outcrops 231
more than 100 m wide, located within the NAF shear zone, of kilometer-width. Due 232
to different deformation markers, two distinct approaches for creeping (volcanic and 233
analogous rocks) and for locked (limestone) sections were used to evaluate the strain 234
values. The techniques are briefly described below including a discussion on the 235
uncertainty of the measurements. These techniques rely on the classical strain analysis 236
procedure well described, for example, in Ramsay (1967, 1980).
237 238
3. Strain measurements methodology 239
240
3.1 Strain measurements along the creeping section (volcanic and analogous 241
rocks) 242
The main parameters of the shear zone are defined in Fig.2a: the kinematic axes a 243
(direction of shear), ab (shear plane) and ac (displacement plane) and the 244
corresponding coordinate axes x, y, and z.
245 246
Boudinage and folding analysis 247
Strain values were extracted from XRF chemical maps relying on the difference of 248
composition of the boudins and the veins relative to the matrix. The quadratic 249
deformed object, and li is its initial length. The quadratic extension is calculated from 251
boudinage or fractured structures: i) the initial length (li) is measured by adding the 252
length of each boudin, and ii) the final length (lf) is the length from the first to the last 253
boudin (Fig. 2b). The quadratic contractional strain is estimated from folded veins 254
with constant length during deformation by measuring i) the final length 𝑙 of the vein 255
and ii) its initial length 𝑙 as the length of an arc along a deformed object using 𝑙 = 256
∑ 𝑑𝑠 = ∫ 𝑑𝑥 1 + (𝑑𝑦/𝑑𝑥) , where 𝑑𝑠 is the length of the section between two 257
successive points and can be written as 𝑑𝑠 = (𝑥 − 𝑥 ) + (𝑦 − 𝑦 ) (Fig.
258
2b).
259
Uncertainty is linked to the ratio between the resolution of the image and the 260
measured lengths. For the boudinage, the total error is the sum of the errors on the 261
measurements of each boudin element that gives the initial length plus the error on the 262
final length. For the folding, it is the sum of the error on initial and final length. In the 263
following, uncertainties are given with the results and are shown as error bars in the 264
figures.
265 266
Volume change due to mass transfer 267
The relative mass change was estimated by comparing the composition of insoluble 268
and soluble minerals between protected (as undeformed as possible) and exposed 269
(deformed) zones at the micro- to meso-scales within a given shear zone (Fig. 2c).
270
The relative mass change is 𝑑𝑚 𝑚 = 𝐼 𝐼 − 1⁄ ⁄ , where 𝐼 and 𝐼 are the 271
concentration of all insoluble minerals in the protected and exposed zones, 272
respectively (Gratier et al., 2013). If the rock density does not change, which is the 273
case in the studied samples, the mass change is equivalent to the volume change. In 274
plane strain, the product of the two principal strain values is related to the volumetric 275
change as follows:
276
𝜆 𝜆 = 1 + ∆, (1)
277
where ∆ is the change in volume (or surface area) (Ramsay, 1967). Because this 278
measurement is based on mineral maps, the accuracy depends on the dimension and 279
on the resolution of these maps. Then, the relative mass change depends on the 280
amount of the insoluble minerals in the protected and exposed zones. An error, 281
defined as a percentage of the calculated volume change or strain, is given in the 282
following. In general, it is smaller than 2%.
283 284
Grain geometry and orientation analysis 285
Mineral phases were identified on the chemical maps and the following geometrical 286
properties of individual grains were measured: the minor and major axes of each grain 287
(rmin, rmax), the grain orientation φg defined clockwise relative to the North, and the 288
coordinates of the center of each grain (xc, yc) (Fig. 2d). In such cases, the standard 289
deviation of the data is calculated and used to estimate the uncertainty.
290 291
Fry method 292
This method is also known as center-to-center method (Fry, 1979, Genier and Epard, 293
2007). The center of every strain marker (e.g. oxide grains) is plotted with respect to 294
the position of all other markers. A plot is built by locating one marker at the origin 295
and by plotting as dots the positions of all other markers. Then another marker is 296
located at the origin and the positions of all other markers are plotted, until all 297
markers have been considered. The result is a cloud of points that contains an empty 298
space (i.e. void) at the origin and which represents the strain ellipse. Strain is 299
quantified using i) the aspect ratio 𝑅 between the elongation axis (a) and the 300
contraction axis (b) of the strain ellipse; and ii) the angle between the x-axis and the 301
elongation axis also defined as 𝜑 . This method is similar to an auto correlation 302
function, which can be used for the same type of object (Heilbronner, 2002;
303
Heilbronner and Barrett, 2014). The normalized center-to-center method (Erslev, 304
1988; Erslev and Ge, 1990) improves the Fry method by taking the grain size into 305
account and was used in the present study. The strain ellipse parameters (𝑅 , 𝜑 ) 306
were measured by superimposing all minerals together (Fig. 2e). The actual values of 307
the principal strain axes were then obtained using 𝑅 = 𝜆 / 𝜆 with 𝜆 = 308
𝑅 (1 + ∆) and 𝜆 = (1 + ∆) 𝑅⁄ . Uncertainty derives from fitting an ellipse 309
to the points along the rim of the void in order to extract (𝑅 , 𝜑 ) values. For best 310
fitting higher weights are given to points that are close to the rim while none of the 311
clustered points remain out of the calculation as suggested by Mulchrone (2013). The 312
angle of the ellipse was fixed using the main orientation of the grains. We evaluate 313
Cleavage angle evolution within the shear zone 316
Evolution of the cleavage orientation within shear zone was used to evaluate the 317
displacement (Ramsay, 1980; Ramsay and Graham, 1970) when integrating the 318
volume change (Fossen, & Tikoff, 1993) at millimeter to decimeter scales on thin 319
sections, and up to macro-scale at meter to hectometer scales. In a shear plane, 320
passive line markers (e.g. cleavage or vein) originally making an angle 𝛼 with the 321
shear zone walls (Fig. 2f) are sheared to make a new angle 𝛼′ (Ramsay, 1980) such 322
that 323
cot(𝛼′) = ( )
∆ (2)
324
High 𝛾 values are very sensitive to 𝑎 angles field measurements: 𝑎 values of 1°, 0.5°
325
or 0.1° lead to 𝛾 values of 57, 114, or 572, respectively (Eq. 2). Consequently, 326
estimating 𝛾 from 𝑎′ is extremely challenging for high strain values, due to the 327
difficulty of measuring 𝑎 with sufficient accuracy when it is smaller than 1°. In the 328
following, the uncertainty of this technique is given as the standard deviation of the 329
calculated strains.
330 331
3.2 Strain measurements along the locked sections (limestone) 332
The strain contraction component was calculated by two techniques. The first one 333
involves identifying veins shifted by stylolites (Gratier et al., 2013) and using ∆𝑙 = 334
𝑙 /tan (𝛽), where ∆𝑙 is the thickness of the layer dissolved by the stylolite, 𝑙 is 335
the distance between the vein walls, and 𝛽 is the angle between the direction of the 336
vein and the stylolite peaks (Fig. 2g). The initial length of the vein is then obtained by 337
adding the shifted length to the final length of the shifted vein 𝑙 = 𝑙 + ∆𝑙, and the 338
quadratic contraction can be calculated. The second technique involves using the 339
maximum amplitude of the stylolite peaks (Toussaint et al., 2018) and 340
applying: 𝜆 (𝜑) = ( ) ( )
( )∙〈 ( )〉, where 𝜆 is the
341
quadratic contraction due to stylolite in their peak direction 𝜑, 𝑁 is the number 342
of stylolites in direction 𝜑, 〈𝐴𝑚𝑝 〉 is the average maximum peak height in 343
direction 𝜑, and 𝐿 is the total length of the sample in direction 𝜑. The strain 344
extension component is measured perpendicular to the vein direction. The final length 345
is defined from side to side and the initial length is the spacing between the veins.
346
Finally, the contraction and extension components are combined and plotted in polar 347
coordinates to define the strain ellipse. Uncertainty is linked to the ratio between 348
resolution of the image and measured lengths. The total error is the sum of the errors 349
on the measurements of shifted veins or stylolites size and on the final length and is 350
given with the results.
351 352
3.3 Relation between shear displacement and finite strain values 353
For simple shear followed by volume change (Fig. 2f), the deformation matrix 354
transforms the undeformed vector (𝑥, 𝑧) (initial state) to a new position after 355
deformation (deformed state) (𝑥 , 𝑧 ): 𝑥
𝑧 = 1 𝛾,
0 1 + Δ ∙ 𝑥
𝑧 where 𝛾 , is the 356
shear strain due to shear and dilation (Fossen and Tikoff, 1993), and the principal 357
strains are:
358
𝜆 , = 1 + 𝛾 + (1 + ∆) ± [1 + 𝛾 + (1 + ∆) ] − 4(1 + ∆) (3) 359
where ∆ is the volume change. Both the aspect ratio between the principal strain axes 360
𝑅(𝛾, ∆) = 𝜆 𝜆⁄ , and 𝜃 (𝛾, ∆) are nonlinear functions of 𝛾 and ∆ (i.e. surfaces in a 361
three-dimensional coordinate system). Such functions are known as 𝑅 − 𝜃 diagrams 362
when plotted together (Fossen and Tikoff, 1993). 𝜃 is the angle between the 363
quadratic extension 𝜆 and shear direction. The total shear displacement 𝑑 is 364
obtained by integrating the shear strain along the z-axis (Ramsay, 1980), (Fig. 2) 365
𝑑 = ∫ 𝛾 ∙ 𝑑𝑧 (4)
366
where 𝑤 is the width of the shear zone, and 𝛾 is the shear strain.
367 368
Figure 2. Strain measurements in shear zones. (a) Kinematic axes with a the direction of 370
shear, ab the shear plane, and ac the displacement plane, and corresponding coordinates axes 371
x,y,z. (b) Example of a boudinage strain marker at meso-scale. (c) Magnification of protected 372
and exposed zones with their mineral composition and strain ellipse. (d) Definition of 373
measured parameters of a single grain: major and minor axes rmin, rmax the aspect ratio Rg and 374
the angle φg, and the location of the center of the grain (xc, yc). (e) Definition of a strain 375
ellipse using the Fry method with major and minor axes a, b and the aspect ratio between 376
them RFry=a/b with the angle φFry. (f) Schematic deformation in simple shear coupled to 377
volume change, with distortion of a unit circle into an ellipse, adapted from (Ramsay, 1967, 378
1980). (g) Contraction evaluated from shifted veins along tectonic stylolite.
379 380
4. Results of finite strain measurements at selected sites 381
The structural strain markers differed from one outcrop to another although all were 382
consistent with dextral slip on the North Anatolian Fault. Therefore, the strain values 383
were extracted using different methods, depending on the particular type of strain 384
markers along the shear zones. For five selected sites (Taskesti, Hamamli, Ismetpasa, 385
Yazioren, Mülayim, Fig. 1), strain measurements at thin section scale are shown in 386
Figures 3, 4, 5, 6, and 7. For each outcrop, the mineral maps, mineral content 387
histograms, Fry and rose diagrams have a similar color code. For the last selected site, 388
Gerede (Fig. 1), strain measurements were integrated at the outcrop and regional 389
scales (Fig. 8). Finally, all strain measurements on the six outcrops are integrated in 390
Figure 9.
391 392
4.1 Taşkesti outcrop (locked section) 393
The Taşkesti outcrop (Fig. 1) exposes the fault zone in massive limestone formations, 394
which are deformed by tectonic stylolites associated with veins as seen in a hand 395
sample (Fig. 3). This sample is well representative of the seven visited outcrops in 396
limestones, which all show several relatively narrow, meter-wide, shear zones with 397
numerous stylolites. The cumulated width of all these narrow shear zones is a few 398
meters at maximum for the whole NAF width. Six thin sections in three orthogonal 399
planes were cut from the Taskesti sample. Two types of tectonic stylolites can be 400
distinguished from the different color of their insoluble residues (either red or black) 401
(Fig. 3a). Several generations of veins can be seen that cross cut each other. Some are 402
sub-parallel to the shear fault plane; others are oblique to it. In places, some of the 403
veins are dissolved and shifted perpendicularly to the stylolite plane, indicating the 404
shortening direction. In other places, some of the horizontal veins are stylolitized, 405
while some vertical stylolites are filled with calcite (Fig. 3b). Figure 3c shows the 406
orientation of maximum amplitude peak directions and stylolite surfaces that are 407
oriented in various directions, with the highest density in the range N70 – N90. It also 408
shows the various orientations of the oblique-to-the fault veins, the most frequent one 409
in the direction N110, and other maximum directions range between N120 and N150 410
(Fig. 3c). This observation is well indicative of the successive generations of 411
stylolites and veins in non-coaxial deformation with alternatively seismic and 412
aseismic deformation.
413
The total strain was measured from stylolites and veins as follows. Firstly, the 414
extension was measured perpendicular to veins from both the hand sample and thin 415
sections (Fig. 3a-b). The strain values obtained from the thin sections were slightly 416
higher compared to the hand sample (Fig. 3d). Secondly, the contraction was 417
quantified by two independent measurements (see section 3.2) using: i) shifted veins 418
and ii) stylolite average amplitude. Both methods gave very similar results with a fit 419
of the strain ellipse with 𝜆 = 1.2 ± 0.1, 𝜆 = 0.9 ± 0.02. The main results from 420
these analyses were that, at the decimeter scale, the calculated mass change was ∆=
421
0.08 ± 0.008, and the shear strain values calculated from the stylolite values (as only 422
the stylolites are representative of the creep process) led to 𝛾 = 0.20 ± 0.05 (Eq. 3).
423
It must be noted that, assuming that the shear fault corresponds to the shear direction, 424
the angle of the ellipse with the fault plane, which is about 30°, could indicate higher 425
shear value of about 𝛾 = 1.15. However, in such semi-brittle deformation, it is 426
difficult to evaluate the orientation of the shear direction, which is not always parallel 427
to the fault.
428
Figure 3. Taskesti outcrop. (a) Limestone hand sample in a horizontal displacement plane 430
(ac, Fig. 2a) in the Taşkesti outcrop deformed by tectonic stylolites associated with two 431
families of veins. Locations of the two thin sections Ta1a1 and Ta1a2 are indicated. A third 432
thin-section, Ta1d, is not shown here. (b) Digitized tectonic stylolites and veins. (c) Rose 433
diagrams representing the orientations of the tectonic stylolite surfaces (red) and peaks (gray), 434
and the first and second generations of veins. (d) Strain ellipse based on stylolite and vein 435
data giving contraction and extension values (black arrows), respectively. The locations of 436
individual veins v1 to v8 are given in (b).
437 438
4.2 Hamamli outcrop (creeping section) 439
The Hamamli outcrop (Fig. 1) exposes serpentinite rocks deformed by dissolution 440
along cleavage surfaces associated with precipitation in veins (Kaduri et al., 2017). In 441
places, the strain was measured by the alignment of iron oxide minerals that 442
agglomerate sub-parallel to the cleavages. The strain was measured along two planes:
443
a shear plane on the fault (ab) and a displacement plane (ac) (Fig. 2a). The results are 444
presented in Fig. 4. The strain values are 𝑅 = 2.3 ± 0.1 and 𝑅 = 3.2 ± 0.2 in 445
the shear fault plane (ab) (Fig. 4a) and in the displacement plane (ac) (Fig. 4b), 446
respectively. Rose diagrams of the iron oxides showed alignments in the strain 447
elongation direction in the displacement plane (Fig. 4b) but were slightly oblique to 448
the strain elongation in the shear plane (Fig. 4a). This result may be due to the effect 449
of episodic oblique displacements along the fault surface. Because it was not possible 450
to find an area that was less deformed than the studied samples, the mass change 451
could not be calculated. A maximum shear strain value in the displacement plane 𝛾 = 452
1.3 ±0.1 was derived from Eq. 3 (Fig. 2f).
453 454
455
Figure 4. Hamamli outcrop. Serpentine rocks show cleavage-foliation and anisotropic 456
clustering of iron oxide minerals in the cleavage. These microstructures were used to evaluate 457
the strain values in two directions. (a) Strain analysis of a fault surface in vertical shear plane 458
(ab), with the rose diagram of Fe oxides, and plots of the normalized Fry using Fe oxides. (b) 459
Strain analysis with the same two methods in a horizontal displacement plane (ac). The shear 460
arrows are indicative of the sense of shear and do not show an accurate orientation of the 461
shear direction, which is not available on the outcrop where the sample was collected.
462 463
4.3 Ismetpasa outcrop (creeping section) 464
boundary with sandstone-shale units. At this site, the deformation in the gouge is 466
associated with tectonic layering leading to more or less fine-grained foliated to 467
granular rocks (F-gouge to G-gouge), Fig. 5a. The foliations are deformed around a 468
rigid clast of residual domain of the initial (undeformed) rock. Figure 5a shows a 469
digitized horizontal thin section, parallel to the displacement plane (ac) of the shear 470
zone, with four XRF maps of silicon on which the Fry-method analysis gave four 471
𝑅 values ranging from 2.7 ±0.1 to 7.5 ±0.4 at centimeter scale (Fig. 5b). It was 472
also possible to calculate the strain distribution at millimeter scale within this thin 473
section from six EPMA maps of mineral content acquired in exposed (the most 474
deformed areas) and protected zones (the less deformed area which is the clast of 475
undeformed initial rock), Fig. 5a. Some of the maps were masked in order to obtain a 476
more accurate evaluation of localized strain and mineral composition in zones of 477
disaggregated but relatively undeformed large quartzite clasts (map 3, Fig. 5a). The 478
relative mass change was calculated excluding magnesite and calcite from the 479
calculation considering that these minerals sealed the porosity of the rock more 480
recently in the deformation process and at least after the process of massive 481
deformation considered here (Kaduri et al., 2017). There was a clear decrease in 482
soluble mineral content from the protected (undeformed) zone to the exposed zones 483
(Fig. 5c). Then, in parallel, the strain was measured on these EPMA maps using the 484
Fry method and 𝑅 values for few representative minerals (albite, anorthite, 485
orthoclase, quartz, smectite). These values were plotted as a function of the relative 486
mass change and showed a near-linear trend (Fig. 5d). The values of 𝑅 converted 487
to quadratic contraction/extension 𝜆 , using the ∆ values (Eq. 1), showed the actual 488
successive change in deformation as a function of the relative mass change (Fig. 5e).
489
Maximum shear strain values are given by maps 2 and 1 with 𝛾 = 1.2 ± 490
0.1 and 1.3 ± 0.1 with 𝑅 = 7.3 and 4.5 and with ∆= – 0.66 and -0.3, respectively.
491
The orientation of the grains in the exposed zones was most often aligned with the 492
local foliation. On the contrary, the protected (near undeformed) zones showed 493
random grain distribution with negligible preferred orientation (Fig. 5f). It must be 494
noted that the estimated grain orientations and the Fry plots are not parallel, because 495
the calculations are not performed at the same scale and because the deformation is 496
heterogeneous around the clast. The Fry plots are obtained from large XRD maps that 497
represent a sampling of the whole thin section (area XRF I, II, III, IV) whereas the 498
grain orientation are obtained from local EPMA chemical maps and are located in 499
zones where the orientation of the deformation is pretty the same (area map 1, 2, 3b).
500 501
502 Figure 5. Ismetpasa outcrop. (a) Thin section in horizontal displacement plane (ac) with two 503
types of gouge layers: foliated and granular gouges (F-gouge and G-gouge), around an 504
undeformed clast of initial rock, with four XRF maps of silicon (I-VI, centimeter-scale) and 505
six detailed EPMA maps of minerals located by numbers (1-6, millimeter scale), each 506
corresponds to a different mineral, given in the legend in (c). (b) Fry method based on the 507
four XRF large maps (I-IV) with their location above and using the minerals in the legend in 508
(c). (c) Mineral content of the mineral maps. (d) Strain measurements using the Fry method 509
is lower than the symbol size). The reference (protected zones) is map 5 (clast). (e) Plot of the 511
calculated quadratic contraction/extension as a function of the relative mass change. (f) Rose 512
diagrams of grain orientations from mineral maps with mean grain diameters (GD). The color 513
code is the same as in (c). The shear arrows are indicative of the sense of shear and do not 514
show an accurate orientation of the shear direction, which is not available on the outcrop 515
where the sample was collected.
516 517
4.4 Yazioren outcrop (creeping section) 518
The Yazioren outcrop (Fig. 1) exposes a shear zone in a mélange of carbonate and 519
volcanic rocks with tectonic layering (rich in alumino-silicates) and carbonate 520
boudinage. A horizontal thin section parallel to the displacement plane (ac) of the 521
shear zone was imaged by XRF in order to segment and measure strain markers.
522
Figure 6a shows the calcium XRF map, which emphasizes carbonate boudinage. The 523
strain measurement included two steps. The first step was to measure the quadratic 524
extension 𝜆 along the preserved boudinage (strain markers 1 to 8 in Fig. 6a), which 525
varied in the range 1.4 ±0.1 - 4.1 ±0.2. The second step was to measure the 526
deformation recorded in the deformed matrix located near the boudins, using the Fry 527
method on maps of elements acquired from EPMA measurements (location m1-m13, 528
in the Al map Fig. 6b). The 𝑅 values varied from 1. 0 ± 0.05 to 4.5 ±0.2. The 529
plot of these two types of measurement showed a good correlation with a fitted linear 530
regression coefficient R2=0.89 (Fig. 6c). The relative mass change was calculated 531
using the EPMA maps by comparing exposed and protected zones. Figure 6d shows 532
an example of these two types of zones: an exposed zone (map13) and a protected 533
zone (pressure shadow contoured with a dashed line in map2) around a calcite clast.
534
From the protected to the more exposed zones, the mineral assemblage showed a 535
reduction in soluble mineral content and a relative increase in insoluble mineral 536
content (Fig. 6e). It is thus possible to calculate the mass change relative to the 537
protected zone (map2) (Fig. 6f). There is a near-linear trend between these strain 538
values evaluated by two different methods, 𝜆 and 𝑅 , and the mass change (Fig.
539
6f). Note that such mass change values were only minimum values since the protected 540
zone was not an initial state but rather a less deformed state as attested by the 541
preferred orientation of the grains (Fig. 6g). Based on these measurements, boudinage 542
values associated with mean ∆ = −0.1 lead to strain ratios ranging from 2 to 19 (Eq.
543
1) and gave a shear strain in the range 𝛾 = 1.3 ± 0.1 to 4.0 ± 0.2 while the Fry 544
method gave a range 𝛾 = 0.5 ± 0.05 to 1.5±0.1 (Eq. 3). These values may be 545
compared with the value of 𝛾 = 1.3 ± 0.1 obtained by the Fry method at the 546
decimeter scale with 𝑅 = 2.8 ± 0.1 (Fig. 6h) and considering a mean value ∆ = 547
0.1. The variation of 𝛾 values at millimetric scale indicated the development of 548
parallel-to-the-cleavage differentiated layers with various associated shear values.
549 550
551 Figure 6. Yazioren outcrop. (a) Calcium map measured by XRF showing boudinage with 552
variation in extensional strain in the horizontal displacement plane (ac). (b) Aluminum map 553
measured by XRF showing the foliation with the locations of the EPMA mineral maps. In (a) 554
and (b) color bars indicate relative content. (c) Correlation between 𝜆 extension of the 555
boudinage and RFry of the nearby matrix (with map-numbers). (d) SEM-BSE image and 556
EPMA mineral maps, showing protected (right) and exposed (left) zones. The color-coding 557
for the minerals is given in the legend of (e). (e) Comparative mineral compositions, the 558
dashed line separates soluble (mobile) and insoluble (non-mobile) minerals. (f) Extension 559
values 𝜆 (from boudinage with numbers) and RFry (from EPMA mineral maps) versus the 560
relative mass change (uncertainty is lower than the symbol size). (g) Rose diagrams of grain 561
orientations from EPMA mineral maps with mean grain diameters (GD) and orientation with 562
respect to the North. (h) Fry diagram at centimeter scale on the whole thin section from XRF 563
data using the minerals in the legend in (e). The shear arrows are indicative of the sense of 564
the outcrop where the sample was collected.
566 567
4.5 Mülayim outcrop (creeping section) 568
In this outcrop (see Fig. 1), the fault crosscuts a block of ophiolitic mélange 569
embedded into schist, marble and metabasite units. Measurements were performed in 570
the schist that constitutes the main fraction of the rock body. Deformation was 571
measured based on the Fry method at different scales with also the use of folded 572
carbonate veins and the statistical analysis of grain clusters. A horizontal thin section 573
parallel to the displacement plane (ac) revealed a shear zone with cleavage - foliation 574
sub-parallel to the fault and crosscut by carbonate veins in all directions (Fig. 7a).
575
The cleavage can be seen on the XRF aluminum map (Fig. 7b). Parallel-to-the 576
cleavage and oblique-to-the-cleavage veins registered only a small part of the finite 577
extension 𝜆 (0°) = 1.1 ± 0.02 , and 𝜆 (90°) = 1.2 ± 0.02 , respectively.
578
Contraction perpendicular to the foliation was measured from folded veins giving 579
𝜆 = 0.45 ± 0.01 from the decimeter-scale thin section (Fig. 7a) to millimeter size 580
EPMA map (Fig. 7d) indicating a minimum shear strain 𝛾 =1.8 ±0.1. The 581
deformation evaluated from the Fry method using XRF aluminum maps had a 582
relatively high value 𝑅 = 8.5 ± 0.4 (Fig. 7c), corresponding to a shear strain 𝛾 = 583
2.6±0.2. No volume change can be calculated here due to the lack of clear evidence 584
of undeformed (or at least less deformed) protected zones, despite cleavage-foliation 585
being visible with alternating phyllosilicate-rich and albite-quartz-rich tectonic layers 586
indicating significant pressure solution mass transfer (Fig. 7f). Nonetheless, the 587
deformation was highly heterogeneous. Using EPMA maps at millimeter size, some 588
strain variations can be measured when using the Fry method in three different zones:
589
the hinge (H) and the limb (L) of an arcuate structure and an intermediate area (I) 590
(Fig. 7d) with 𝑅 ranging from 1.4 ±0.07 to 4.8 ±0.2 (Fig. 7e). Such strain values, 591
lower at millimeter size than at decimeter scale, implies the existence of some highly 592
localized zones with very high strain values at the sub-grain size. Within these zones, 593
phyllosilicates were oriented sub-parallel to the shear zone with embedded residues of 594
feldspar minerals and Fe oxides (Fig. 7f) and such fine-grained thin structures were 595
probably not taken into account well enough by the Fry method.
596
This problem was investigated by a statistical analysis of the Al grain clusters. More 597
precisely, relying on the detailed observations of the grain orientation and associated 598
Fry ellipses (Fig. 7d, f, e), the distribution of the orientation of elongated grain 599
clusters containing aluminum was fitted with two normal probability density 600
functions. At the decimeter scale (Fig. 7b) the cleavage-foliation is overall sub- 601
parallel to the fault at the boundary of the sample. In such ductile context, this fault is 602
presumably parallel to the shear direction and, by measuring the angle 𝜃 between the 603
line normal to the fault and the long axis of elongated grain clusters, the amplitude of 604
shear strain is evaluated: the angle between the long axis of elongated grains 605
(assimilated to cleavage plane) and the shear direction (a) is 𝛼′ when 𝛼 = 45° (initial 606
cleavage) in Eq. 2. This calculation corresponds to the wider normal distribution in 607
Fig. 7g that gives an angle 𝛼′= 1° associated with 𝛾 value of 57. However, at the 608
millimeter scale (Fig. 7d, f) we described two types of shear bands: those with grains 609
oblique to the shear direction and those with grains sub-parallel to this shear direction.
610
In order to better constrain the value of the 𝛼 angle in the shear bands with grains 611
sub-parallel to the shear direction we choose to concentrate our analysis in these 612
bands (with angles ranging from -5 to + 5° with the shear direction). Doing so we 613
avoid the possible perturbing effect of the grains oblique to the shear. This calculation 614
corresponds to the narrower normal distribution in Fig. 7g that gives an angle 𝛼′=
615
0.3°. This value of the 𝛼 angle is associated with 𝛾 value of 190. However, 616
uncertainty on the angle between the normal distribution of the elongated grains 617
cluster and the shear displacement is probably higher than 1°, leading to 𝛾 values 618
ranging from 40 to > 200. The coexistence of such a wide range of strain values, the 619
uncertainty of their evaluation and their use to calculate the aseismic displacement is 620
discussed in section 5.2.
621 622
Figure 7. Mülayim outcrop. (a) Calcium map in the horizontal displacement plane (ac) 624
measured by XRF with a folded calcite vein (ptygmatic vein) shown between arrows along 625
dashed white lines. (b) Aluminum map in the same sample indicates sub-parallel cleavage- 626
foliation rich in phyllosilicates. In (a) and (b), color bars indicate relative content. (c) Fry 627
diagram based on XRF-maps (decimeter scale, 7a-b) using the minerals in the legend in (d):
628
anatase and mica. (d) Mineral maps with fold separated by masks in the hinge, the 629
intermediate area and the limb with different cleavage angles (locations are given in Fig. 7a).
630
(e) Fry diagram based on the selected zones of the mineral map with increasing values of RFry
631
from hinge to limb (millimeter scale) using the minerals in the legend in (d). (f) Mineral map 632
in a highly localized deformation zone with sub-parallel to the shear displacement micas and 633
embedded fine-grained quartz. (g) Orientation distribution of clusters of grains containing Al 634
(Fig. 7b) fitted by two normal probability density functions (PDF), with the mean diameter of 635
the grain clusters containing aluminum (GCD). Parameters of the two normal distributions are 636
<α'> the mean angle with the fault (F) and σ' the standard deviation: <α'> = 1° and 0.3° and 637
σ'= 42° and 2.5° for the wide and narrow distributions, respectively.
638 639
4.6 Gerede outcrop (creeping section) 640
The Gerede outcrop (Figs. 1 and 8b) is an abandoned quarry at the western end of the 641
Ismetpasa creeping segment (Hussain et al., 2016; Cetin et al., 2014). It is located one 642
kilometer south of the present-day active fault zone with several other major strike- 643
slip faults in between mostly buried under Quaternary sediments (Kaduri et al., 2017).
644
The quarry presents a continuous 150 m-wide outcrop (Fig. 8a) that reveals a network 645
of 5 to 30 cm thick clay-rich soft gouges with parallel-to-the-fault cleavage-foliation 646
(Fig. 8a, 8c, 8d). These faults cross the volcanic units of the Galatia massif 647
(Adiyaman et al., 2001), which are a mélange of dacite, andesite and trachy-basalts 648
(Wilson et al., 1997). These different units are difficult to distinguish. In the gouges, 649
they have been transformed by deformation coupled to low temperature 650
metamorphism (Kaduri et al., 2017). In the damage zones, they have been highly 651
damaged with fractured blocks and dense networks of vertical veins in all directions 652
(Fig. 8e). Such sealed fractures were related to episodic inflow events of carbonate- 653
rich fluids that were associated with the successive earthquakes that affected this 654
section of the NAF (Kaduri et al., 2017). Moreover, the deformation was 655
heterogeneous with more or less rigid fractured blocks surrounded by foliated zones 656
with intense deformation displaying a braided fault pattern at all scales, from thin 657
section (Fig. 5a) to regional scale (Fig. 8a). The size of such rigid blocks decreased 658
drastically in gouges and in zones with cleavage sub-parallel to the shear fault but the 659
boundaries of such zones were undulated and their widths varied along strike (Fig.
660
8a-c-d). In thin sections, only some islands of volcanic rocks can be seen between the 661
networks of veins. Consequently, it was not possible to evaluate the strain and shear 662
values using particle distribution as in the other studied outcrops.
663
However, it was possible to evaluate the shear values from the angles between the 664
cleavage and the shear fault planes, similar to what was done at microstructural scale 665
for the Mülayim outcrop (Fig. 7g). Such angles can be measured either directly in the 666
field or in thin sections. It was not possible to measure the variations of these angles 667
along a continuous path over the entire 150 m of the outcrop. However, the damage 668
zones appeared to be more layered in areas near the gouges than away from them 669
(Fig. 8c-d). Thin sections showed that, in these areas near the gouges, the cleavage 670
orientation was sub-parallel to the shear fault plane (Fig. 8f-g-h), whereas away from 671
the gouge the cleavage was oblique to the fault (Figs. 8e), as in other outcrops (Fig.
672
5f, 6g). Such change of cleavage orientation being rather sharp, it was possible to 673
evaluate in the field the width of the shear zones characterized by parallel-to-the-fault 674
cleavage with an uncertainty of about 10%. Such shear zones include all the gouges 675
plus damage zones around them that have the same structural aspect as that seen on 676
width of the quarry. The ratio between this cumulative width of shear zones with 679
parallel-to-the-fault cleavage and the total 150 m width of the outcrop may be 680
evaluated and is equal to 0.04 ±0.004. This ratio can be considered as a representative 681
value at a more regional scale (100-1000 m). The shear strain values associated with 682
cleavage sub-parallel to the fault were not easy to evaluate and will be discussed in 683
section 5.2. These strain values and the width of the shear zones with fault parallel 684
cleavage were used to estimate the total fault displacement accommodated by creep 685
during geological times in section 5.3.
686 687
688 Figure 8. Gerede outcrop. (a) Photograph of the 150 m-wide outcrop showing the fault 689
network and the location and widths of zones (shaded in white) with parallel-to-the-fault 690
cleavage/foliation. (b) Geological map with the trace of the main fault in the outcrop (dashed 691
line), Q=Quaternary, Pl3=Pleistocene conglomerate sandstone, M7=Miocene agglomerate 692
tuff, T7=Eocene andesite basalt, MaPa=Cretaceous limestone, Sna=Cretaceous limestone, 693
JKk1=Jurassic limestone. (c) Zone with parallel-to-the-fault cleavage and soft gouge in the 694
upper part of the quarry: see location in (a). (d) Zone with parallel-to-the-fault cleavage and 695
soft gouge in the lower part of the quarry (not seen in a). (e) Element map of Ca using XRF 696
with typical network of calcite - dolomite veins parallel and perpendicular to the cleavage.
697
The blue perpendicular-to-the-fault vein is a late magnesite vein. (f) Parallel-to-the-fault 698
cleavage in a thin section of soft gouge. (h) Parallel-to-the-fault cleavage in a thin section 699
near the gouge (see location in c). (g) Parallel-to-the-fault cleavage in SEM image with 700
dolomite and magnesite veins parallel to the fault and to the cleavage - foliation (layers of 701
clinochlore and smectite).
702 703
5. Discussion 704
Several parameters are needed to calculate the displacement associated with creep 705
processes in the NAF shear zone: the quadratic extension and contraction strain 706
values, the volume change, that allow calculating the shear strain when data of all 707
outcrops are combined (Fig. 9), and the width of the shear zone that allows 708
calculating the shear displacement (Eq. 4). The evaluation of these parameters is 709
discussed below, as well as their uncertainty. In section 5.1 we discuss the volume 710
change in the creeping sections and its effect on shear displacement. In section 5.2 we 711
discuss the strain and width evaluation. We evaluate the total displacement in locked 712
sections. We discuss the effect of heterogeneous deformation in creeping sections and 713
we propose a model of two normal shear strain distributions. Finally, the modeling of 714
the total aseismic displacement along the creeping sections is discussed in section 5.3.
715 716
5.1 Volume change evaluation: uncertainty and effect on shear displacement 717
Here we discuss the relationship between strain and mass-volume change because 718
volume change is an important parameter of the relation between total strain and shear 719
strain (Eq. 3). These relations are discussed based on a “strain-ratio/shear- 720
strain/volume-change” diagram (Fig. 9). An horizontal plane strain in 2D is assumed, 721
thereby implying homogeneity along the y-axis (Ramsay, 1980; Heilbronner and 722
Barrett, 2014; Fossen and Cavalcante, 2017). For the same measured strain ratio, R, 723
the higher is the volume change the lower is the associated shear strain (Fig. 9). For 724
example, for R=10, the shear strain ranges from 𝛾 = 2.9 to 𝛾 = 2.0 for a volume 725
change ranging from ∆ = 0 to ∆ = −0.5, respectively (Fig. 9). For higher strain ratio, 726
such as R=10000, when cleavage become sub-parallel to the shear zone at less than 727
0.5° angle, the shear strain ranges from 𝛾 = 100 to 𝛾 = 72 for a volume change 728
ranging from ∆ = 0 to ∆= −0.5, respectively (Fig. 9). Despite such variations of 729
strain for such large variation from ∆ = 0 to ∆= −0.5, it does not change the order of 730
magnitude of the shear strain values. It is worth noting that the calculated mass 731
change is a relative value. It is obtained by comparing a deformed “exposed” area 732
with a “protected” area which is either undeformed (in the best case) or which is just 733
less deformed. This is the reason why we also evaluated the deformation of the 734
protected zone from grain geometry and orientation analysis. Moreover, as the rock 735
density does not significantly change, mass change is equivalent to volume change.
736
representative of all visited sites along the NAF shear zone (Fig. 1): Taskesty in 738
limestone, Hamamli in serpentine and the four other sites (Ismetpasa, Yazioren, 739
Mulayim, Gerede) in various rocks with volcanic origin.
740
All the sites in limestone show several narrow parallel-to-the-fault shear zones of 741
decimeter width with numerous stylolites. The Taskesti sampling is well 742
representative of all these shear zones for which the deformation near the fault 743
measured at decimeter scale shows no significant volume change with ∆ = 0.08 (Fig.
744
3). Observations (Fig. 3) always show clear evidence of alternating seismic (fractures) 745
and aseismic (stylolites) processes (Kaduri et al., 2017). However, the aseismic part 746
remains very low, with mean quadratic values of 𝜆 = 1.2 ± 0.1 and 𝜆 = 0.9 ± 747
0.02 leading to a low shear strain value 𝛾 = 0.2 ± 0.05, a value not significantly 748
affected by uncertainties on volume change.
749
At the Hamamli serpentine site, it was not possible to calculate a volume change as no 750
area less deformed than any other was found. This difficulty is a general problem of 751
volume change calculation in serpentine where heterogeneities are rare (Andreani et 752
al., 2005). In Mulayim, the same absence of protected zone prevented us from 753
evaluating the volume change. The corresponding strain values at these sites are then 754
plotted in Fig. 9 assuming no volume change.
755
At the other sites, volume change was measured by comparative chemical analyses 756
(Fig. 5 & 6) at millimeter scale. These calculations assume that in the deformed areas 757
the soluble minerals dissolve and are removed. A minor part of these soluble 758
minerals, as some feldspars, were transformed into newly crystallized phyllosilicate 759
minerals rather than removed, however, this does not significantly alter the 760
calculations (Kaduri et al. 2017). There is a clear correlation between strain and 761
volume change indicating that the same pressure solution mechanism is responsible 762
for the deformation with the development of parallel-to-the-cleavage tectonic 763
layering. When the protected zone is in an undeformed state, as in Ismetpasa (Fig. 5), 764
the volume change varies from one layer to another from 0.22 ±0.005 to more than 765
0.66 ±0.01. When the protected zone is slightly deformed, as in Yazioren (Fig. 6), 766
the variations are lower and the uncertainty is much higher. This pattern of successive 767
perpendicular-to-the-compressive-stress layers with various compositions is typical of 768
a pressure solution self-organized process that has been reproduced experimentally 769
(Gratier et al., 2015). This observation was made in the 25 investigated outcrops in 770
volcanic and analogous rocks of the NAF creeping shear zone. It indicates a very slow 771
ductile and aseismic deformation (Gratier et al., 2013). In order to extrapolate this 772
quantitative result to the whole shear zone, some complementary geological 773
observations are needed. No traces of recrystallization of the soluble minerals (quartz 774
and feldspar) are found in the shear zone, so it is likely that the whole shear zone was 775
developed with a decrease of volume at least during the early time of its development.
776
However, later in the NAF development, numerous carbonate veins crosscut some 777
damage zone as seen in Gerede (Fig. 8e). As such carbonates need to come from 778
outside since they were not present in the initial rocks, they may have contributed to a 779
local increase of volume. Consequently, in order to evaluate shear strain values 780
representative of the entire NAF shear zone in volcanic and analogous rocks (Fig. 9) 781
we used a median regional value of the volume change with a large uncertainty, ∆ = 782
−0.20 ± 0.15.
783
Finally, the possibility of much higher volume changes, that we may have missed, 784
must be discussed because large volume change values below ∆ = −0.8 have a large 785
effect on the calculation of shear strain (Fig. 9). Common observation is that volume 786
change associated with pressure solution is limited to about ∆ = −0.7. It is due to 787
several effects especially the progressive decrease of the soluble mineral content and 788
their progressive isolation in the soft matrix of insoluble species that render their 789
stress-driven dissolution more and more difficult (Gratier, et al., 2013). Moreover, 790
there is another effect in shear zones that limits the volume change, which is linked to 791
the rotation of the cleavage. Pressure solution initiates along solution cleavage planes 792
near rigid objects (Fig. 5a) and develops progressively further away at 45° from the 793
shearing direction at the beginning of the shearing process. At this stage, the shearing 794
displacement contributes to the tectonic layering process with volume decrease.
795
However, at a later stage with the rotation of the cleavage that becomes sub-parallel to 796
the shear zone, the displacement is less and less dependent on the volume reduction 797
perpendicular to the cleavage. When the cleavage is sub-parallel to the shear zone 798
(Fig. 7f), only the effect of the perpendicular-to-the-shear zone compaction 799
contributes to the dissolution. At this stage, large sliding along parallel-to-the-fault 800
cleavage must switch to grain boundary sliding. This mechanism can always be a 801
pressure solution process at grain size scale in order to accommodate relative grain 802
